Checklists good – this message has been spread through all tech conferences for the last several years, so I didn’t expect much from this book. It’s the slightly more fleshed-out (but also just inflated) version of the original essay. If I hadn’t heard all the good conference talks about pilot checklists (for example https://rixx.de/blog/andrew-godwin-you-have-control/), I’d have rated this book higher.
While the anecdotes that extend the essay to book length are mildly repetetive, I still enjoyed reading them. While the upshot is well-known: people routinely mistakes despite knowing better, checklists are proven to work beyond people’s wildest dreams. Examples are mostly from medical contexts, with some civil engineering, professional cooking and military applications in between. But apart from the core message, there were some details that stood out to me, including when and where checklists are applicable and how to write and introduce them.
Checklists improve baseline performance. Lack of expertise is not the problem, rather the opposite: Complexity and volume of our knowledge are beyond our ability to live up to it reliably and safely. When the average ICU patient requires around 180 actions per day, even the mistake rate of 1% starts to look pretty bad. (The parts of the book that were about ICUs were an interesting read given the current pandemic: For example, people who spend 10 days on a ventillator have a 6% chance to develop an infection that will be deadly in 45% of cases.)
So what do checklists add? A bunch of things, and remembering the actual steps to take is not even the most important one. For example, checklists give you a reason or an excuse to follow the steps, when self-pressure and peer pressure would have you skip them. They give you a legible reason to check other people’s work, including people higher up in the hierarchy (for example, nurses double-checking a surgeon’s preparation). Though, remember: not every problem is suited to checklists – checklists are for simple problems (as in: there is a clearly defined recipe/set of steps to follow, given a set of starting conditions). We want people to get the simple things right consistently while also leaving room for craft and judgement.
Introduction of checklists, as all new procedures, needs to convince everybody involved: the people using them, and ideally somebody with sufficient power from within management who commits to running the introduction. One technique that was highlighted a lot was the necessity of the team in question talking to each other, even if just to tell each other their names and responsibilities. It’s one of those tricks that stops people from feeling like bystanders, and makes them take initiative towards a positive outcome.
Writing checklists is a skill. When checklists are vague, imprecise and annoying, they will be ignored and they lead people to turn off their brains. Good checklists are precise, efficient, and easy to use in difficult situations. They don’t spell everything out – they just remind of critically important steps (rule of thumb: 5-9 steps). To begin, checklists must define a clear pause point, at which it is supposed to be used. Then, you have to decide for a DO-CONFIRM list (do stuff, then pause and run through the list) or a READ-DO list (carry out tasks as you check them off). The wording needs to be simple, exact, and in familiar professional language. The list should fit on one page and be free of any clutter and unnecessary colour. Make them very fast to check off.
And then, you have to iterate on real world tests, because you never get it right the first time. Investigating actual failures and successes is critical, as unpleasant as it may seem. And remember to put a publication date on every new version - no checklist lasts forever.